Topic: Incentive of selfish mining under volatile block rewards (Read 350 times)

That plurality is the problem. The article also says the selfish miner would mine multiple blocks and keep them private. But unless they have >51% of the hashrate, their chain will always be shorter than the actual chain when they keep building on it.

That's true. But will there really be a financial gain? The way I see it, what authors explain is basically a 51% attack (replacing already mined blocks and reversing some transactions). That can crash the price and whatever the selfish miner was hoping to earn would be worth a lot less. Meaning it would act as a disincentive breaking any kind of coalition between malicious miners, effectively lowering their total hashrate therefore making the repetition of the attack impossible.

Satoshi assumes there's a financial incentive to keep nodes honest:

If there's a situation where being dishonest gives a financial gain, I totally expect human greed to take over.

I agree with your highest hashrate assumption, but I thought selfish mining would function with a real-life assumption in practice.

For selfish mining, we're assuming that the attacker is able to generate blocks and they're able to withhold the blocks before broadcasting them. If another block appears at the same height, it should push their own blocks and depending on how much time they've withheld, they would've done a certain amount of work on expanding their own chain rather than the rest. As long as they have an edge over the other chain, then there is a chance that the rest of the miners would work on your chain rather than theirs.

I recall the contentious point as largely about the profitability and feasibility back in 2013. However, I'm also going to go on a limp to assume that certain assumptions has changed 13 years (!) since then, particularly on the assumption on propagation delay and miner's connectedness.

Skimming through the article I keep remembering the same "assumption" we have always had in Bitcoin. Let me quote Satoshi's whitepaper:

So when the authors talk about "Selfish Mining" and how the miner would create an alternative private chain, in order to get that accepted by the rest of the network replacing the public chain, the selfish miner has to have the chain with the most amount of work which requires higher hashrate than the rest of the network combined. Otherwise the selfish miner's chain will become orphaned not the public chain that is mined by the majority of the hashrate that also should have the most amount of work.

P.S. I should also point out that the authors of this article don't understand something as basic as the difference between orphaned blocks and stale blocks!

That's a good observation about the interplay between miner competition, block reward, and volatility.  You're right, the dynamic of miners vying for inclusion in the next block is a significant factor.  The intensity of this competition, influenced by things like hash rate distribution and network congestion, can contribute to price fluctuations indirectly , a sudden spike in mining difficulty, for example, might correlate with a temporary price drop as miners adjust.
However, attributing direct volatility solely to this competition is an oversimplification.  While selfish mining strategies like the one you described could theoretically manipulate block generation times, the cost of doing so is the risk of being detected .significantly it  limits their profitability.  The block subsidy is a factor, but other market forces like macroeconomic conditions, regulatory news, adoption rates, and speculative trading are far more influential drivers of Bitcoin's price volatility.
The existence of a block subsidy is important, not just because it directly affects profitability but also because it incentivizes miners to secure the network in the first place.  If the only reward was transaction fees, it would be much harder to attract and retain the necessary mining power. 

It will also be more expensive for selfish miners as difficulty adjustments incur losses under the fixed reward model. Same as undercutting before and after difficulty adjustments which could pose a strong effect on mining attack on long and short term.

True. And this will also mean reward per block will no longer be fixed. And honest miners will repeatedly get their block orphaned  and excluded from the canonical chain. This volatility in block reward will alter the profitability of various mining strategies. Unlike I'm the case of fixed reward model where miners are incentivized to mine the longest chain where there will obviously be no benefit orphaning previous blocks.

Why would a client do that? They'll have to pay more in transaction fees, and it takes longer to be confirmed. If their transaction is confirmed in the earliest block, and even if a reorg happens after that, their transaction is still finalized faster than if it was confirmed in a later block. I don't see any benefit for the Bitcoin user who just wants to make a transaction.

Maybe my post was not clear enough, so I will rephrase it: I think people will see in advance, that the reward for the block number N is higher, than the reward for blocks around it. Which means, that if it will be mandatory, to split the same amount between many blocks, then the problem will be solved. Which means, that people will come up with some soft-fork proposal, where you will have it splitted into "X coins for now, Y coins for later". And as long, as nothing like that is deployed yet, you can safely come up with any proposal, where any proportions can be used, to fix it. Also, as far as I know, some sidechains like RSK are fee-based, and they already use some fee-balancing algorithms, so the mainnet solution can be based on something similar.

Exactly. Each and every transaction has a field called "locktime". Which means, that clients could even estimate some "reorg risk", based on network data, and put a proper locktime in created transactions, to balance fees. Currently, by default, many clients use "zero" or "last block number", but it can be changed. And then, if a given transaction will lock some fees into a future block, then miners will have an incentive, to produce enough blocks, to unlock those fees.

No. It's the opposite: a high fixed block reward disincentivizes miners from doing "selfish mining".

This isn't about block rewards, it's about transaction fees. A low block reward and high transaction fees in only some of the blocks can encourage miners to reorg the chain and replace the valuable block by their own.

Does this mean that halving can actually be seen as some sort of future fix for the problem of selfish mining? Because as the number of halving events begin to increase the reward per block decreases meaning by the next halving, block rewards would become way smaller and thus kind of less valuable meaning miners would begin to depend more on fees until all the coins have been successfully mined.

By next halving the block reward would decrease by 1/2 meaning it would now be a question of choosing between 0.1BTC and 0.25BTC and next halving it would be between 0.1BTC and 0.125BTC meaning as time passes the selfishness towards fees could drop.

It was brought up a while back, but it was dismissed after a while given that the opportunity cost is too high currently. Malicious re-orgs is generally not good, I don't think a few more confirmations would be particularly reassuring. Miners don't have any incentive to try to game the system right now, but if it becomes a prominent and persistent issue, then I think it warrants some concerns.

Anyways, I think there's also an issue of pushing miners to be more centralized. If the system favours larger miners in this case, then small miners are probably not inclined to mine alone. This point is kind of debatable given how mining is already centralized.

I only skimmed through the article, but your example explains it faster. My question is: is this really going to be a problem? We'll just have to wait for more confirmations. At the moment 1 confirmation is enough to be fairly sure the coins aren't going anywhere, and that might change. But other than that, Bitcoin will be fine. Unless miners go wild and reorg more than a few blocks.

There's already a pretty strong incentive to do so, and as far as I know all miners "sort" transactions to include only the most valuable ones.

Of course he could not have known how ASICs would develop, but the difficulty adjustment system he created can still handle it. That means it was designed to scale up to any hashrate miners can throw at it.

Moore's law would've made it pretty obvious and increased participation would've also indicated that mining would get increasingly more competitive and industralized. GPUs, CPUs were fairly consistent with their performance gains. FPGAs and ASICs were also obviously coming along and PoW pretty much guarantees that these would happen, it is actually by design. The existence of these were intended.

Actually, the original client didn't include a block limit which meant that this problem never really existed (well, miners can impose it if they want, but that's out of the question). Having a fee market and scarcity of block space actually solves the issue of miners being underpaid from fees. There's tons of tradeoffs in Bitcoin and no perfect model will ever exist.

Not exactly a cause for concern. Miners pours in tons of upfront capital and contracts, acting against the benefit of all it bound to incur fairly significant losses. Any schemes which favours those with the most resources will end up being less decentralized than desired. No going around it.

Current model is not really the best model. I believe that Bitcoin is not the final product of Satoshi because he hasn't considered many things that would come with increased usage. I believe he didn't know that we would have so powerful GPUs, CPUs, SSDs, ASICs and so on.
I know it doesn't directly answer the quote but I want to say that Satoshi's model of Bitcoin isn't perfect and that's the issue. There is a competition because everyone wants to get their transactions confirmed as soon as possible. Imagine you want to buy something with Bitcoins and you can't wait, you try to get your transaction confirmed ASAP to receive the order confirmation email.

Yes, Bitcoin mining is losing decentralization and I think that that's an alarming case.

Miners are already incentivised to prioritize higher fees over lower fees, that is by the design of the network. As it stands now, Bitcoin is already far less decentralized with 70% of the network centralized around 4 mining pools. They can always collude to artificially inflate the fees at any point in time, just that orphaning blocks is an entirely different ball game and the incentive would be drastically different.
I doubt that we would want to interfere with the free market forces in the fee market. There would be more problems created than solved in this case. It's quite difficult to incentivise miners to immediately broadcast whatever blocks that they have mined since the consensus function would have to be modified significantly.

Ironic as it seems, it is a beautiful thing for this to be noticed because it shows advancement of research findings to man. Exploit can be find on almost anything with time and that is known, now before it actually become a real pain for decentralization and the Bitcoin network, a feasible solution will have likely been implemented, because a problem known is a problem that can be solve.

I read through the abstract and I can say a fixed rate might indeed be better to avoid this, although not disrupting the  fact that an improved solution might even come before then.

You do have a solid point. The potential for selfish mining to become kinda more viable in the transaction-fee era actually raises topics  worth discussing about the long-term security and probably possible flaws of decentralization on the Bitcoin network. Because If miners can profitably manipulate the blockchain by orphaning high-fee blocks it could actually lead to a concentration of power among large mining pools undermining the network's decentralized nature.

Moreover, this could create a stronger incentive for miners to prioritize high-fee transactions over lower-fee ones, which can even lead to increased transaction fees and decreased network accessibility for smaller miners. These are more like flaws that exist in a decentralised system because even if power is relatively shared it can still be abused by people who hold huge amounts just like mining pools with tremendous hash rates.

In the paper, it is demonstrated that difficulty adjustment is currently discouraging selfish mining, because the selfish miner has to incur a significant loss during initial epoch, either because his blocks are getting orphaned, or because he is not mining for that period at all to decrease difficulty (and therefore incurs the real cost of not mining).

But in the transaction-fee era, difficulty adjustment does not carry the same weigh. Losses during the initial epoch may be mitigated or even eliminated, depending on the variable fee rate. For example, if median block reward is 0.1 BTC, and due to network congestion, the most recent block paid the miner 0.5 BTC, it might be a profitable strategy for a selfish miner to orphan it, because claiming 0.5 BTC with just one block is more profitable than mining five.

I actually glanced through the pdf file, and from a technical Bitcoin perspective the writeup highlights an interesting dynamic related to block reward and mining strategy but most negative ones are still not feasible.

Aside from the fact that bitcoin block reward is designed to decrease over time, with the halving event occurring approximately every 4 years,  As block reward decreases transaction fees become a more significant component of mining in general meaning smaller sats will have more value as time passes.

Personally I think one reason selfish miners won't be able to take advantage of those vulnerabilities is because the bitcoin network has the ability to adjust it's difficulty depending on the last mining duration. This means if they extend block  generation time then next block's  difficulty will be reduced So it would become more difficult for them to extend block generation time further.

I stumbled across this paper that was published a month ago, describing mining strategies on the post fixed block reward era. I found it interesting for discussion, and I share it: https://arxiv.org/pdf/2411.11702.

The competition of people trying to get their transactions included in the next block, being the primary factor that determines block reward, is what causes the volatility. If a selfish miner removes honest miners' blocks from the canonical chain, it can extend the block generation time, and make the competition more intense, which will increase his profitability. The existence of real cost (block subsidy) does not allow this strategy to be profitable at the moment.